Copyright 2008 Year IEEE. Reprinted from IEEE ECTC May 2008, Florida USA.. This material is posted here with permission of the IEEE.

Transcription

1 Copyright 28 Year IEEE. Reprinted from IEEE ECTC May 28, Florida USA.. This material is posted here with permission of the IEEE. Such permission of the IEEE does not in any way imply IEEE endorsement of any of Institute of Microelectronics products or services. Internal or personal use of this material is permitted. However, permission to reprint/republish this material for advertising or promotional purposes or for creating new collective works for resale or redistribution must be obtained from the IEEE by writing to

2 Angled High Strain Rate Shear Testing for SnAgCu Solder Balls Chai TC a, Yu DQ a, John Lau a, Zhu WH b, Zhang XR b a Institute of Microelectronics, A*STAR (Agency for Science, Technology and Research), 11, Science Park Road, Science Park II, Singapore b United Test & Assembly Center Ltd. (UTAC) 5, Serangoon North Ave 5, Singapore taichong@ime.a-star.edu.sg, Fax: (65) Abstract A high speed shear test of solder ball at different shear angles has been developed and investigated for SnAgCu solder balls. The objective of the test is to introduce vertical loading component in the high speed shear test and to study intermatallic response of different solder balls. The shear test is performed at three angles of attack e. g. 3, 5 and 1 degree. The basic shear tester allow high speed shear tool movement of up to 4 mm per second, and capture the loaddisplacement curve experienced by the shear tool. Using this setup, high-speed shear characterization has been carried out on microbga sample with Sn3Ag.5Cu and Sn1Ag.5Cu solder on Ni/Au pad finish. In this paper, the effect of angle shear at high speed on these SnAgCu BGA balls is presented, with the details of load-displacement response and failure mode for these solder balls. Generally all the samples show changes in ductile-tobrittle failure as shear speed increases. The angled shear test lead to down shift of ductile-brittle transition to occur at lower shear speed with larger shear angle. It is found to improve sensitivity in revealing brittle failure for solder ball characterization. Introduction Due to the susceptibility of drop impact damage on solder interconnection, the popularity of portable products has led to a greater concern on solder joint performance under high strain rate loading conditions. At the present, high strain rate behavior of solder joint has been evaluated using high speed shear test on component level or board level drop impact testing. While the board testing is time consuming and laborious, the correlation of component level high speed shear testing on BGA solder ball is still under investigation by different researchers [1, 2]. One of the main issues faced by this approach is the limitation of test: that it requires component mounted on board for testing, and little info about the peeling or tensile loading to the component under test can be obtained from the test. Such a test is difficult to be implemented in the component manufacturing environment where component level ball integrity test are preferred. Today most of manufacturers are still performing the classical ball shear test and some have equipped with ball pull test. However both these technique have their limitations. In reality, solder ball interconnection is subjected to combined shear, tensile and peeling stresses. Hence a realistic assessment of solder ball integrity should incorporate such loading component simultaneously in the same test. Nevertheless, shear test is still considered to be a preferred method for its convenience and ease of implementation [3-6]. For these purposes, we have developed a novel high speed shear test method that that allow high strain rate testing of solder ball at different shear angles. The objective is to introduce simultaneously vertical and horizontal loading to the ball under test in a single ball shear test. Such a mixed mode loading is useful for solder ball integrity assessment as many realistic applications are under mixed mode condition. For demonstration and proof of this concept, we have implemented the test method with three angles e. g. 3, 5 and 1 degree. A commercial available basic shear tester allows high speed shear tool movement of up to 4 mm per second, and to capture the load-displacement response of the shear tool. The test method is made to ensure the point of contact between shear tool and solder ball for all the three shear angle will be consistent on the solder ball is achieved. Using this setup, high speed shear characterization has been performed on 2 types of solders, namely Sn3Ag.5Cu and Sn1Ag.5Cu on microbga package with Ni/au pad finish. Test specimen and test procedure The microbga sample is shown in Figure 1 (a), where the total I/O numbers is 54 with solder ball size and pitch at.4 mm and.8 mm respectively. The solder ball is consisted of Sn3Ag.5Cu and Sn1Ag.5Cu. Solder pad material is Cu with sold-mask-defined (SMD) configuration, and coated with electrolytic NiAu finishing. During shear experiments, the load-displacement responses of the shear tool are captured for different shear speed and shear angle. After the testing, the sheared samples are subjected to microscope and SEM inspection for failure mode identification. To determine a meaningful shear speed range for the experiment, some initial shear tests have been performed to ensure the ductile-brittle transition will be covered in the shear speed experiment. In this case, the tool speeds range from 5 mm/s to 25 mm/sec has been selected /8/$ IEEE Electronic Components and Technology Conference

3 Shear force (g) Zero degree degree 12 Figure 1: Samples for ball shear test (a) micobga package; (b) Solder ball cross-sectional profile Shear force (g) Shear angle and shear speed on load-displacement curve Figure 2 shows the shear force (g)-displacement (µm) curves from experiment at various shear speeds for Sn3Ag.5Cu samples. The curves plotted in Figure 2(a) (b) (c) and (d) present the shear test under, 3, 5 and 1 degree respectively. From these graphs, some trends can be found. As shearing speed increases, the displacement is decreasing steadily, while maximum shear force increased. This is inline with the general expectation of strain hardening effect of solder material. For low shear speed below 1 mm/s, the changes of displacement are insignificant in both displacement and peak load. As the shear speed increased, there is an increase in the ramp rate of shear force with drastic changes in displacement part. The sharp decreasing of shear displacement is an indication that the solder ball is going through from ductile failure to brittle failure. In between the two failure mode, where some irregular displacement has occurred, the transition of ductile shear to brittle shear is considered to have occurred. Based on the above observation of the load-displacement curves, the shear speed for which different failure mode have occurred is summarized in Table 1, including the shear speed in which ductile-brittle transition is considered to have occurred. This result provides a clear picture of the changes taking place for different shear angle. Generally brittle failure tends to initiate at lower shear speed as shear angle increases. This include down shift of ductile-brittle transition to lower shear speed. Despite small increase in shear angle, the result indicates an improvement in the sensitivity of angled shear test in revealing brittle failure than horizontal shear test. Shear force (g) Shear force (g) degree degree Figure 2: Typical loading curves for different shear speed and different shear angles for Sn3Ag.5Cu solder balls Electronic Components and Technology Conference

4 Shear speed (mm/sec) Ductile failure Transition Brittle failure Zero degree 1,5,7,1, 15 25,35 3,5,7 3 degree 1,5,7,1, 7, 15 25,35 3,5 5 degree 1,5,7,1, 3 5,7 15,25, 35 1 degree 1,5,7 1,3 7,15, 25,35 Table 1: Ductile-brittle transition as function of shear angle and shear speed Characterization of failure mode The sheared samples were inspected under optical microscope and SEM for its failure mode after shear testing. For convenient purpose, the failure modes are classified under a few categories namely bulk failure, bulk-imc partial failure, IMC failure, IMC + partial pad peel failure, and pad peeling. Typical image of these failure modes are shown in Figure 4. Figure 3 shows a direct comparison of load-displacement curve extracted for different shear angle for shear speed around the ductile-brittle failure transition region. The effect of shear angle on the load-displacement response described earlier can be clearly observed here. Mode 1: Bulk failure Mode 2: Bulk-IMC partial failure For 3 mm/s degree 3 degrees 5 degrees 1 degrees Mode 3: IMC failure Mode 4: IMC + pad peeling For 5 mm/s For 7 mm/s degree 3 degrees 5 degrees 1 degrees degree 3 degrees 5 degrees 1 degrees Figure 3: Comparison of loading curve as function of shear angle Mode 5: Pad peeling Figure 4: s observed from shear test Figure 5 shows the compilation of failure mode as function of shear speed and shear angle for the Sn3Ag.5Cu solder on Ni/Au finish. The results match well the loaddisplacement curve and once again the ductile-brittle transition is showing a down shift in shear speed as shear angle increased. The result from higher shear angle is clearly showing improved sensitivity in revealing IMC brittle failure. However, it is also observed that at higher shear angle, e.g 1 degree, the occurrence of IMC+partial pad peeling has increased as compared to the lower shear angle samples. We believe that the pad peeling is an indication of weaker pad to core substrate interface integrity. The increase in angle shear is able to reveal such weakness in the sample, subjected to the shear speed. At the same time, it is competing with the bulk- IMC mixed failure depends on which is weaker. As the strain rate further increased to the higher shear speed region, the brittle IMC failure become more apparent Electronic Components and Technology Conference

5 1% 3 Degree shear 9% 8% 7% 6% 5% 4% 3% 2% Pad peeling IMC+pad peeling IMC Bulk-IMC Bulk % Ag 1% Ag 1% % a) 3 degree shear Shear speed (mm/sec) Shear speed (mm/s) a) 3 degree shear 1% 5 Degree shear 9% 8% 7% 6% 5% 4% 3% 2% Pad peeling IMC+pad peeling IMC Bulk-IMC Bulk % Ag 1% Ag 1% % Shear speed (mm/sec) Shear speed (mm/s) b) 5 degree shear b) 5 degree shear 1% 9% 8% 7% 6% 5% 4% 3% 2% 1% % Shear speed (mm/sec) Pad peeling IMC+pad peeling IMC Bulk-IMC Bulk c) 1 degree shear Figure 5: Compilation of failure mode from shear test Degree shear Shear speed (mm/s) c) 1 degree shear Figure 6: Comparison of ductile-to-brittle transition for different shear angles for SAC35 and SAC15 samples 3% Ag 1% Ag Comparison of SAC35 and SAC15 For comparison to Sn3Ag.5Cu, the similar angled high speed shear test was also performed on Sn1Ag.5Cu solder prepared on the similar package and pad finish. Figure 6 shows the compilation of failure mode for the 2 solder types. The results show SAC35 has lower resistance to occurrence of brittle IMC failure compared to SAC15. The result is consistent for all the 3 type of angle shear test. It is interesting to note that large shear angle, e.g. 1 degree, is showing better sensitivity in revealing the transition to brittle failure for the SAC35 as compared to SAC15. We suspect that the higher angle shear test may be able to bring out more subtle aspect of the solder ball integrity compare to normal horizontal shear test. Finite element analysis of angled high speed shear A further verification is also performed by using 3D Finite element modeling of ball shear test to investigate the effect of shear test at different angle. This approach can provide more detail information on the stress distribution in the solder ball and ball/cu pad interface during the test. Solder and Cu pad materials are considered as elasto-plastic material, and solder mask and core are elastic. ANSYS/Mechanical and ANSYS/LS-Dyna are used for the static and dynamic simulation respectively. Only static analysis results are presented here. When shear tool move at constant speed over solder ball, there is a reaction force activated from solder ball when both come into contact. Figure 7 shows the shear and normal force of the central contact node on the shearing ram. It is seen that the normal force from shear ram increases as shear angle Electronic Components and Technology Conference

6 increases, while the effect shear angle on shear force has insignificant change. Point 2 Node shear force (N) Node normal force (N) Angle5 6 8 Angle1 Load Step Angle Angle Load Step Angle Angle3 Angle5 Angle1 Figure 7: Node shear and normal force in static shearing Figure 8 shows the comparison of 3D distribution of normal stress and shear stress in the solder material just above Cu pad between zero and 5 degree shear angle, respectively. It can be seen that both stress patterns showed slight difference, while the normal stress magnitude for 5 degree increase significantly from zero degree. a) Normal stress ( ) b) Normal stress (5 ) Point 1 Normal Stress (MPa) Shear Stress (MPa) Normal Stress 3 Normal Stress 5 Normal Stress 1 Normal Stress Distance along diameter (mm) Shear Stress 3 Shear Stress 5 Shear Stress 1 Shear Stress Distance along diameter (mm) Figure 9: Line stress distribution at solder/cu pad interface extracted along point 1 and point 2. Figure 9 shows the comparison of normal stress and shear stress distribution along the Cu pad/solder interface extracted from the middle cross-section plane of solder ball as function of different shear angle. Once again the result shows a clear trend of increasing normal stress as function of shear angle. This line plot provides a convenient way of looking at stress distribution at the Cu pad/solder interface although it is not necessarily a true reflection of the original 3D stress distribution. Nevertheless, it is suffix for illustration of the normal and shear stress component as the shear angle change. c) Shear stress ( ) d) Shear stress (5 ) Figure 8: Stress distribution in solder ball above Cu pad Conclusions and recommendation High speed shear characterization has been carried out on 2 types of SnAgCu solder on microbga with shear angle of 3, 5, and 1 degree. In addition, 3D finite element modeling has been performed for stress distribution of ball shear test when subjected to change in shear angles. Some important results and recommendation are summarized in the following. 1) Ball shear test at higher shear angle improved the sensitivity of revealing brittle IMC failure of solder ball. 2) The improvement of the angle shear test has been shown in both load-displacement response of solder ball as well the failure mode of the samples Electronic Components and Technology Conference

7 3) The ductile-brittle transition of Sn3Ag.5Cu and Sn1Ag.5Cu solders on microbga package with Ni/Au pad finish has been obtained for different shear speeds and different shear angles. It is found that a. Sn3Ag.5Cu is more prone to brittle failure compare to Sn1Ag.5Cu. b. For Sn3Ag.5Cu, IMC failure was initiated at 7 mm/sec for 3 degree angle shear, and it was reduced to 1 mm/sec for 1 degree shear. c. For Sn1Ag.5Cu, IME failure was initiated at 25 mm/sec for 3 degree angle shear and it was reduced to 7 mm/sec for 1 degree shear. 4) Analysis of the angle shear test showed the peeling stress and normal stress increased in the solder when the shear angle is increased. The increased in peeling and normal stress in the solder ball is believed to be the reason for the improved sensitivity in revealing brittle failure in the angled shear test. 5) A shear test that provides simultaneous shear and pulling stress to the solder ball has been developed and it is recommended for use in solder ball integrity, reliability and IMC characterization purpose. Acknowledgments This work is the result of a project initiated by the 8 th IME Electronic Packaging Research Consortium (EPRC VIII), the members of which are Advanced Micro Devices (Singapore) Pte Ltd, BOC Gases Pte Limited, Ibiden Singapore Pte Ltd, Infineon Technologies Asia Pacific Pte Ltd, NEPES Corporation, United Test and Assembly Center Ltd, MicroCircuit Technology (22) Pte Ltd, Cookson Semiconductor Packaging Materials A Division of Cookson Singapore Pte Ltd, Institute of Microelectronics, Singapore Institute of Manufacturing Technology, Institute of High Performance Computing and Institute of Materials Research & Engineering. The authors are grateful to Mr Andrew Peh of Dage (SE Asia) Pte Ltd for the use of high speed shear equipment, and students Mr Chia CY, Mr Ding Ting, Ms Waiyan and members of EPRC VIII Project 1 and their colleagues in their respective RI in A*STAR who had contributed and made this work possible. References 1. K. Newman, BGA Brittle Fracture Alternative Solder Joint Integrity Test Method, Electronic Components and Technology Conference (55 th ECTC), May 25, p EH wong, YW Mai, R Rajoo, KT Tsai, F Liu, SKW Seah, CL Yeh, Micro Impact Characterization of Solder Joint for Drop Impact Application, Electronic Components and Technology Conference (56 th ECTC), May 26, p E. Kaulfersch, S. Rzepka, V. Ganeshan, A. Muller and B. Michle, Thermal, Mechanical and Multi-Physics Simulation Experiments in Microelectronics and Micro- Systems, EuroSime 27, April 27, p J. Y. H. Chia, B. Cotterell, T. C. Chai, Materials Science and Engineering A 417 (26) X. J. Huang, S. W. Ricky Lee, C. C. Yan, 22 Electronic Components and Technology Conference, p J. W. Kim, D. G. Kim, S. B. Jung, Microelectronics Reliability 46 (26) D. Q. Yu, C. M. L. Wu, D. P. He, N. Zhao, L. Wang, J. K. L. Lai, J. Mater. Res., Vol. 2 (25) M. Amagai, M. Watanabe, M. Omiya, K. Kishimoto, T. Shibuya, Microelectronics Reliability 42 (22) W. Liu, N. C. Lee, JOM (27) Electronic Components and Technology Conference